knitr::opts_chunk$set(echo = TRUE, warning = FALSE, message = FALSE)
library(tidyverse)
library(here)
library(broom)
library(ggtext)
# Time series packages
library(tsibble)
library(feasts)
library(lubridate)
library(fable)
knitr::include_graphics("/Users/matthuy/Desktop/Rprojects/ESM 244/esm244_a4_spatialdata_timeseries_textwrangling/falls.png")
An aerial view of Willamette Falls and the former Blue Heron paper
mill. The Oregonian
Located about 10 miles south of Portland, Oregon on the Willamette
River, Willamette Falls is the largest waterfall in the Northwestern
United States. Despite a complicated industrial past, the falls have
been a native salmon and lamprey fishery for thousands of years and see
many more species of fish pass through since the first fish ladder was
constructed in 1885. The Oregon
Department of Fish and WIldlife maintains a fish counting station at
Willamette Falls, allowing biologists to monitor populations over
time.
knitr::include_graphics("/Users/matthuy/Desktop/Rprojects/ESM 244/esm244_a4_spatialdata_timeseries_textwrangling/ladder.png")
Aerial View of Willamette Falls. 2010. Escapement
Estimate of Adult Pacific Lamprey at Willamette Falls. by Baker,
C.,
J. Graham, Confederate Tribes of the Warm Springs Reservation.
Oregon Department of Fish and Wildlife. 2010. p. 4.
For this time series analysis we will use adult fish count data,
recorded from 2001-01-01 to 2010-12-31. This data includes daily
observations of 13 different fish species that pass through the
Willamette Falls fish ladder. This data was shared by and accessed from
Columbia
River DART (Data Access in Real Time), part of the Colombia Basin
Research project at University of Washington’s School of Aquatic &
Fisheries Management.
### read in fish count data
fish_df <- read_csv(here("data/willamette_fish_passage.csv")) %>%
janitor::clean_names() %>%
mutate(date = as.Date(date, format = "%m/%d/%y")) ###change date col from haracter to date class
fish_df[is.na(fish_df)] <- 0 ### replace NA's with 0
fish_ts <- as_tsibble(fish_df, key = NULL, index = date) %>% ### change data frame to tsibble
select(date, coho, jack_coho, steelhead) ### select date and species of interest
### create static time series graphs
coho_plot <- ggplot(data = fish_ts,
aes(x = date)) +
geom_line(aes(y = coho),
color = "darksalmon") +
labs(x = " ",
y = " ") +
theme_minimal()
jcoho_plot <- ggplot(data = fish_ts,
aes(x = date)) +
geom_line(aes(y = jack_coho),
color = "azure3") +
labs(x = "Date",
y = "") +
theme_minimal()
steel_plot <- ggplot(data = fish_ts,
aes(x = date)) +
geom_line(aes(y = steelhead),
color = "darkseagreen") +
labs(x = "Date",
y = "",
caption = "**Figure 1:<span style = 'color:black;'> Count of adult passage for <span style = 'color:darksalmon;'>coho, <span style = 'color:azure3;'> jack coho, <span style = 'color:darkseagreen;'> and steelhead salmon <span style = 'color:black;'> through the<br> Willamette Falls fish ladder. (01-01-2001 to 12-31-2010)**") +
theme_minimal() +
theme(plot.caption = element_markdown(hjust = -1, color = "black",
size = 10, halign = -0.004))
### arrange in a panel with cowplot
cowplot::plot_grid(coho_plot, jcoho_plot, steel_plot,
labels = c(" Coho ", "Jack Coho", "Steelhead"),
vjust=-0.5,
hjust = -0.2,
ncol = 1) +
theme(plot.margin = unit(c(1, 0, 0, 0), "cm"))
All three species follow clear seasonal trends, with coho and
jack coho passage overlapping their seasonality.
Overall increasing trend in Coho salmon populations, especially
in the last two years. Steelhead populations appear to be slightly
decreasing over time. There is no apparent overall trend in jack coho
populations.
2008 stands out as an outlier in the jack coho passage data;
there were nearly double the amount of jack coho salmon that year
compared to any other year.
coho_ts <- fish_ts %>%
select(date, coho)
jcoho_ts <- fish_ts %>%
select(date, jack_coho)
steel_ts <- fish_ts %>%
select(date, steelhead)
coho_season_plot <- coho_ts %>%
gg_season(y = coho,
alpha = 0.8) +
theme_minimal() +
labs(x = "",
y = "") +
scale_color_gradientn(labels = c("2000", "2002", "2005", "2007", "2010"),
colors = c("lightsalmon", "salmon2", "tomato3", "indianred4", "salmon4"))
jcoho_season_plot <- jcoho_ts %>%
gg_season(y = jack_coho,
alpha = 0.8) +
theme_minimal() +
labs(x = "",
y = "") +
scale_color_gradientn(labels = c("2000", "2002", "2005", "2007", "2010"),
colors = c("gray90", "gray70", "gray50", "gray30", "gray10"))
steel_season_plot <- steel_ts %>%
gg_season(y = steelhead,
alpha = 0.8) +
labs(x = "Month",
y = "",
caption = "**Figure 2:<span style = 'color:black;'> Seasonal adult passage for <span style = 'color:darksalmon;'>coho, <span style = 'color:gray50;'> jack coho, <span style = 'color:darkseagreen;'> and steelhead salmon <span style = 'color:black;'> through the<br> Willamette Falls fish ladder. Lines represent individual years from 2001 to 2010.**") +
theme_minimal() +
theme(plot.caption = element_markdown(hjust = 1, color = "black",
size = 10, halign = -0.004)) +
scale_color_gradientn(labels = c("2000", "2002", "2005", "2007", "2010"),
colors = c("paleturquoise1", "paleturquoise3", "darkseagreen1", "darkseagreen3", "darkseagreen4"))
cowplot::plot_grid(coho_season_plot, jcoho_season_plot, steel_season_plot,
labels = c(" Coho ", "Jack Coho", "Steelhead"),
vjust=-0.5,
hjust = -0.2,
ncol = 1) +
theme(plot.margin = unit(c(1, 0, 0, 0), "cm"))
Coho and Jack Coho adult passages peak from September to
November.
Steelhead adult passages are more spread out than coho and jack
coho, happening from January to August and peaking around June and
July.
These season plots appear to support our observations in Figure 1
that coho salmon populations have increased over time (darker colors
indicate more recent years), while steelhead salom populations have
decreased (darker green colors indicate more recent years). Jack Coho
populations appear relatively stable, aside from an unusually high year
in 2008.
coho_ts_yearly <- coho_ts %>%
group_by(year = year(date)) %>%
summarise(total = sum(coho))